Water Storage of the Central Amazon Floodplain Measured with GIS and Remote Sensing Imagery

Abstract

Interferometric processing of a Space Shuttle–based swath of synthetic aperture radar (SAR) data collected over the central Amazon floodplain reveals one-day decreases in water levels ranging from 1 to 11 cm in mid-October 1994. The decreases, plotted in a geographic information system (GIS), do not appear to be related to water-body size. Instead, the drops correlate with distance between measured location and main river channel. Because the flow routes connecting observed water bodies to a river channel are convoluted, stream network extraction algorithms operating within a GIS are ideally suited to identify flow-path distances. High-resolution digital elevation models (DEMs) are not available for the Amazon Basin; instead, extraction algorithms are applied to a basin-wide mosaic of raster-based SAR amplitude data. SAR amplitudes mimic topography with weak radar returns over open water lying in low elevations and strong returns over non-flooded, higher elevations. As flow-path distance increases, there is a general decrease in the floodplain water level drops. This functional relationship is used to extrapolate the swath of interferometrically based observations beyond their ∼15-percent coverage of the central Amazon floodplain. If the extrapolation is correct, the decreases in water levels sum to a storage change of 4,600 m³/s from an area of 11,800 km² across the floodplain during midrecessional flow. This value is about 30 percent less than the storage change estimates previously derived from flood routing. The difference likely results from the model-based assumption that water-level fluctuations across the floodplain are equivalent to those in the main river channel. The interferometric observations suggest that this equivalence does not appear to be correct for mid-October 1994.

Key Words:

• floodplains
• flow networks
• GIS
• interferometric SAR

Acknowledgments

This research was funded by NASA's Solid Earth and Natural Hazards program (NAG5-9055). I thank my colleagues Tom Dunne, Laura Hess, John Melack, Leal Mertes, and Larry Smith for their comments and continued collaboration. I also sincerely appreciate the efforts of Bruce Chapman, Ake Rosenqvist, and the entire Global Rain Forest Mapping (GRFM) team toward the construction of Amazon Basin wide JERS-1 SAR mosaics.

Notes

Note: Measuring method indicates the flow-path extraction algorithm. Floodplain areas vary with the locations of watershed divides. See text for details.

Note: At 99-percent probability (i.e., α= 0.01, or 1 percent chance of a type I error), the critical value for failure in all tests is 2.33.